Tuesday, April 7, 2009

Playing Baseball With the Monte Carlo Method

This article in the New York Times about using computer simulations to model scenarios in baseball games is really interesting. In essence, rather than try to come up with an ubercomplicated statistical model to predict outcomes, some gies just write a computer simulation program and then model like 100 or 1,000 or even 10,000 entire seasons in order to smooth out the stats and so forth. Now obviously all this depends on how good your simulator is, but since they've been around since the 1950s, I'm guessing they've made some headway in the field.

The most interesting part about this is that the guy who created one of the more well respected simulators is now employed by the Red Sox. I've got to say I'm really impressed by the Red Sox approach to the game. They also employ the guy who invented sabermetrics, which tries to accomplish the same thing as these simulators, but looks at the problem from a different perspective. I'm confident that all of this research affects the types of players that are currently on their roster, but I wonder how much of this actually gets onto the field in terms of strategy and lineups and such. Anyway, whoever decided to hire Theo and these other gies at the Red Sox was a foreward thinking dude.


I like math

1 comment:

Open Bar said...

This is one of the main reasons the Red Sox are gonna be really good for a long time. They have all these brilliant people figuring things out, added to a massive financial pool. It's the kind of organization I wish the Mets would outright copy. But no, we give Luis Frigging Castillo a 4-year deal. Dammit, now I'm all riled up.